Fingerprints and palm prints formed by numerous ridge lines in a curved stripe pattern have been used as a means for verifying a person for a long time. In particular, a matching that uses a latent print remaining in a crime scene, and the like is an effective investigation means. In many police agencies, a fingerprint matching system using a computer is installed. Respective feature points (also referred to as Minutiae) of a fingerprint image registered in a database and the latent print collected at the crime scene, and the like are compared to specify a person corresponding to the latent print. End points and bifurcations of the fingerprint ridge line (fingerprint minutiae) are often used as the feature point used in the fingerprint matching.
However, the quality of the fingerprint image for the latent print is low, and usually contains a great amount of noise. Thus, automatic extraction of the feature point using the computer has limits. The feature point calculated by the computer is often not used thoroughly even in the police agency installed with the fingerprint matching system. In this case, an examiner with specialized knowledge often carries out the operation of manually (by hand) inputting the feature point (end point, bifurcation) to an image for the latent print. Alternatively, the examiner may add or delete a feature point with respect to the fingerprint image, on which the feature point automatically extracted by the computer is reflected, to set the feature point. The addition of the new feature point and the deletion of the feature point by the examiner are referred to as a manual correction input (Minutia edit, Minutia plot) of the feature point.
Patent Literatures 1 to 4 disclose a technique for accurately extracting a feature point from a fingerprint image and a technique for reducing the task necessary for the input task of the feature point, and the like. Specifically, Patent Literature 1 discloses a fingerprint feature extracting device that reduces the task of checking the number of ridge lines and the correction task, which are essential when an editing task of the feature point of an unclear fingerprint, in particular, the latent print is carried out. Patent Literature 2 discloses a ridge line direction extracting device, and the like that can accurately extract the ridge line direction even in a direction unstable region. Patent Literature 3 discloses an image processing device that removes a curved stripe pattern noise from the curved strip pattern image and emphasizes the ridge line of the fingerprint in interest, and the like. Patent Literature 4 discloses a fingerprint feature correction system that assists the correction task of the fingerprint image for the latent print, and the like.